data_mining_flashcards

Question 1

Front

Answer 1

Back

Question 2

What is data mining?

Answer 2

The process of discovering patterns and extracting useful knowledge from large datasets.

Question 3

What are the two main types of learning?

Answer 3

Supervised: Labeled data, prediction tasks (Regression, Classification). Unsupervised: Unlabeled data, pattern discovery (Clustering, Association).

Question 4

Give examples of data mining applications

Answer 4

Healthcare, fraud detection, marketing, finance, recommendation systems.

Question 5

What are common techniques to handle missing data?

Answer 5

Drop rows/columns with missing data. Imputation (mean, median, mode).

Question 6

What is normalization vs. standardization?

Answer 6

Normalization: Rescales values to [0, 1]. Standardization: Centers data with mean = 0, std = 1.

Question 7

What is one-hot encoding used for?

Answer 7

To convert categorical features into binary columns.

Question 8

What is the purpose of EDA?

Answer 8

To explore, visualize, and summarize data to find patterns and outliers.

Question 9

What is a boxplot?

Answer 9

A visualization tool to show data spread, quartiles, and detect outliers.

Question 10

What does a correlation matrix tell us?

Answer 10

The relationships (positive/negative/none) between numerical variables.

Question 11

What is the formula for linear regression?

Answer 11

Y = b0 + b1X + ε.

Question 12

What is logistic regression used for?

Answer 12

To predict binary outcomes using the sigmoid function.

Question 13

What are common evaluation metrics for classification?

Answer 13

Accuracy, Precision, Recall, F1 Score, ROC-AUC.

Question 14

How does KNN work?

Answer 14

It classifies a point based on the majority class of its k-nearest neighbors.

Question 15

Name two distance metrics used in KNN.

Answer 15

Euclidean, Manhattan.

Question 16

What is the impact of increasing k in KNN?

Answer 16

Larger k reduces overfitting but can oversmooth boundaries.

Question 17

What is entropy in decision trees?

Answer 17

A measure of impurity in a dataset.

Question 18

What is pruning in decision trees?

Answer 18

Reducing tree size to prevent overfitting.

Question 19

What criteria are used for splitting in decision trees?

Answer 19

Gini impurity, Entropy, Information Gain.

Question 20

What is bias in machine learning?

Answer 20

Error from overly simplistic models (underfitting).

Question 21

What is variance in machine learning?

Answer 21

Error from sensitivity to training data (overfitting).

Question 22

What is PCA used for?

Answer 22

To reduce dimensionality while retaining important features.

Question 23

What are eigenvectors and eigenvalues?

Answer 23

They define the principal components in PCA.

Question 24

How does a random forest work?

Answer 24

It uses bagging to create multiple decision trees and combines their outputs.

Question 25

Why is random forest better than a single decision tree?

Answer 25

It reduces overfitting and improves accuracy.

Question 26

What is the goal of SVM?

Answer 26

To find the optimal hyperplane that maximizes the margin between classes.

Question 27

What are the assumptions of Naive Bayes?

Answer 27

Features are conditionally independent given the class.

Question 28

What are the steps in KMeans clustering?

Answer 28

1. Initialize k cluster centers. 2. Assign points to the nearest cluster. 3. Update cluster centers.

Question 29

How does GMM differ from KMeans?

Answer 29

GMM uses probability distributions, while KMeans uses centroids.

Question 30

What is the Apriori algorithm used for?

Answer 30

To identify frequent itemsets in market basket analysis.

Question 31

What are Support

Answer 31

Confidence

Question 32

What are content-based and collaborative filtering?

Answer 32

Content-based: Uses item features. Collaborative: Uses user interactions.

Question 33

What is survivorship bias?

Answer 33

Bias caused by focusing only on successful cases.

Question 34

How is NLP used in data mining?

Answer 34

Text analysis for tasks like sentiment analysis and machine translation.